Central Venous Catheters Coated with Minocycline and Rifampin for the Prevention of Catheter-Related Colonization and Bloodstream Infections

Methods

Study Sample

Our study was conducted simultaneously at five university-based hospitals in the Texas Medical Center in Houston: The University of Texas M.D. Anderson Cancer Center (518 beds), Veterans Administration Medical Center (1050 beds), Hermann Hospital (600 beds), Ben Taub General Hospital (580 beds), and The Methodist Hospital (904 beds). The study began on 1 September 1994 and ended on 27 March 1995.

Hospitalized patients 18 years of age or older who required a triple-lumen polyurethane central venous catheter at a new insertion site were asked to participate. We excluded pregnant women, patients who were allergic to rifampin or tetracycline, patients with dermatitis or a burn over the insertion site, and patients for whom the anticipated duration of catheterization was less than 3 days. All patients gave informed consent.

Randomization

All catheters were triple-lumen, polyurethane, 7 French, and 20 cm long (Cook Critical Care, Bloomington, Indiana). The coated catheters were pretreated with tridodecylmethyl-ammonium chloride and then coated, 18 hours later, with minocycline and rifampin. The levels of minocycline and rifampin on the external and internal surfaces of coated catheters before insertion, as determined by high-performance liquid chromatography, were 139.3 µg/cm and 13.9 µg/cm, respectively. Control catheters were untreated and uncoated. All catheters were gas sterilized and placed in identical trays, and each tray was assigned an identification number. The trays were then randomly assigned into blocks of six: three with coated catheters and three with control catheters. Each block of trays was placed in boxes by Cook Critical Care, and the boxes were shipped to the five hospitals. When a patient was determined to be eligible, a tray was removed from the box (trays were removed one at a time, in sequential order from top to bottom), and that catheter was used for the patient. The catheter identification number was recorded on a data entry form and on the patient's medical chart; neither the patient nor the clinician who inserted the device knew which catheter (coated or uncoated) had been used.

Catheter Insertion and Care

Study catheters were inserted into the subclavian vein, internal jugular vein, or femoral vein of patients who had no other indwelling catheter. Study catheters were not exchanged over guidewires. Maximal sterile barrier precautions were taken, including use of a sterile gown, sterile gloves, full sterile drapes, a mask, and a cap. At the time of catheter insertion and at each dressing change, the insertion site was cleaned with chlorhexidine gluconate (at The Methodist Hospital) or 10% povidone-iodine scrub (at all other hospitals). In each case, the preparation was applied to the skin for 2 minutes before catheter insertion. The insertion site was then covered with sterile gauze and taped securely. The insertion site was inspected every 72 hours (during a dressing change) for evidence of infection, such as erythema, purulence, swelling, or tenderness over the catheter. During follow-up, the following information was obtained for all patients: site of catheter insertion; dates of catheter placement and removal; occurrence of difficulties and violations of aseptic technique during insertion or removal, if any; reason for using the catheter (chemotherapy, total parenteral nutrition, administration of blood products, or a combination of these reasons); type of dressing; and reason for catheter removal. In addition, clinical data were obtained on underlying disease, neutrophil and platelet counts, antibiotic therapy administration, other therapeutic interventions administered during the period of catheterization, and the presence or absence of fever and infection during catheterization. The catheter remained in place until it was no longer needed; until a specific event, such as catheter-related infection, necessitated its removal; or for 28 days, whichever occurred first.

Microbiological Methods

Quantitative Cultures of Central Venous Catheters

The entire catheter was removed aseptically, and 4-cm segments were cut from the catheter tip and the subcutaneous section. These segments were semiquantitatively cultured by using the roll-plate method; the same segment was then quantitatively cultured by using the sonication method [12-14]. Organisms recovered by either method were fully identified according to standard microbiological methods. Coagulase-negative staphylococci were classified as gram-positive cocci in clusters that produced catalase but not coagulase and were categorized according to species by using the Staph-Ident System (Analytab Products, Plainview, New Jersey). All hospitals used the same methods for culture.

Skin Cultures

To determine whether bacteria became resistant to the antibiotics that coated the study catheters, skin samples obtained from the insertion site were cultured at the time of insertion and within 24 hours after catheter removal, as described elsewhere [15]. Organisms recovered from the insertion site were fully identified by using standard microbiological methods.

Antimicrobial Resistance

We used the modified Kirby-Bauer technique to test the antimicrobial activity of the catheters coated with minocycline and rifampin against all organisms isolated from indwelling coated catheters at the time of catheter removal [16]. The zones of inhibition against staphylococci cultured from coated catheters were compared with those of uncoated catheters. The minimal inhibitory concentration (MIC) of minocycline hydrochloride (Lederle Laboratories, Pearl River, New York) and rifampin (Ciba-Geigy Corp., Summit, New Jersey) against staphylococcal organisms that colonized the catheter tip, subcutaneous segments, and adjacent skin insertion sites of the coated catheters was determined. A microbroth dilution method was used to determine the MIC in accordance with guidelines established by the National Committee for Clinical Laboratory Standards [17].

Definitions

The definitions adopted for our study were proposed by the Centers for Disease Control and Prevention [18]. Colonization of a central venous catheter was defined as 1) the isolation from either the tip or the subcutaneous segment of 15 or more colony-forming units of any organism by the rollplate technique or 2) isolation of more than 1000 colony-forming units of any organism by the sonication technique. Catheter-related bloodstream infection was defined as the isolation of microorganisms from the bloodstream (blood was obtained through venipuncture, not through the catheter) of a patient who had concurrent clinical manifestations of sepsis and no source for the bloodstream infection other than the vascular catheter. In addition, the catheter had to be colonized with the same organism (same species and same antibiogram). To confirm the diagnosis of catheter-related bloodstream infection, DNA molecular typing done using pulse-field gel electrophoresis was performed on organisms that were of the same species, had the same antibiogram, and were isolated from the catheter and blood during the period of catheterization. Patients were considered to have fever if the oral body temperature was greater than 38 °C. Neutropenia was defined as a polymorphonuclear count of fewer than 1000 cells/mm³. Thrombocytopenia was defined as a platelet count of fewer than 100 000 cells/mm³.

Molecular Typing

Molecular typing was performed by using pulse-field gel electrophoresis. Identical organisms with similar DNA profiles that were isolated from a segment of the colonized catheter and from the bloodstream confirmed the diagnosis of catheter-related bloodstream infection. However, a mismatch did not rule out such a diagnosis because catheter colonization may be polyclonal within a species. The data on bloodstream infections were analyzed with and without the confirmation of molecular typing. Bacterial plug preparation, restriction enzyme digestion, gel preparation, and electrophoresis were done according to the methods described by Maslow and colleagues [19].

Statistical Analysis

The analyses of catheter-related colonization and bloodstream infection included all patients who were enrolled in the study and had their catheters cultured by quantitative methods. All statistical tests were two tailed. Categorical variables were analyzed by using the Fisher exact test, and 95% exact CIs were calculated by using StatXact software (Cytel Software Co., Cambridge, Massachusetts). Continuous variables were compared by using the Mann-Whitney test. Times to catheter-related septicemia in the two study groups were estimated according to the Kaplan-Meier method and were compared by using the exact log-rank test (StatXact software). The incidence rates of catheter-related bloodstream infection were computed per 1000 days of catheterization and were compared by using a binomial (exact) test for incidence rates (Stata software, Computing Resource Center, Santa Monica, California). The multivariate logistic regression model was used to simultaneously correct for the effect of multiple variables and their interactions on colonization. An exact multivariate logistic regression model using LogXact software (Cytel Software Co.) was performed to correct for the effect of multiple variables and their interactions. Factors and interactions that were significant at a P value of 0.25 or less in a univariate analysis were entered into both analyses and tested for an independent effect through a backward technique. The limit for entering and removing variables in these models was a P value of 0.05 or less.

Industry Role

No company, including Cook Critical Care, was directly or indirectly involved in the design of the study, collection or analysis of the data, or the decision to submit the report for publication.

Results

Patient Characteristics

A total of 298 venous catheters (147 coated and 151 uncoated) were inserted in 281 patients. Thirty patients had 32 catheters removed without subsequent culture, and 30 patients had catheters in place for less than 3 days (this finding was not anticipated before randomization). The two groups were similar with respect to host factors, therapeutic interventions during the study period, and hospital sites where the catheter was inserted (Table 1). The analysis of outcome included the 30 catheters that were removed after less than 3 days of insertion.

Characteristics of Central Venous Catheters

The two study groups were similar for location of catheter insertion, difficulty of catheter insertion, and reason for catheter removal (Table 2). In addition, both groups had catheterization for a median of 6 days (Table 2). The 32 catheters that were not cultured and thus were excluded from the outcome analysis were equally distributed between the two groups (17 coated and 15 uncoated catheters); host factor and catheter characteristics in these catheters were similar to those listed in Table 1 and Table 2.

Colonization of Central Venous Catheters

Catheter colonization was determined in all 266 catheters (removed from 251 study patients) that were quantitatively cultured, including the 30 catheters that remained in place for less than 3 days. Colonization was detected in 36 (26%) of the 136 uncoated catheters and in 11 (8%) of the 130 coated catheters (P < 0.001; relative risk, 3.13 [95% CI, 1.66 to 5.88]; risk difference, 18% [CI, 7.9% to 28.8%]). Similar trends were seen across the five participating hospitals. In addition to catheter coating, the significant factors in the univariate analysis were female sex, intravenous hyperalimentation, insertion site, insertion in the intensive care unit, and use of systemic antibiotic therapy and interleukin-2 (P ≤ 0.25). These variables were entered into a multivariate logistic regression model, and five independent protective factors were revealed: catheters coated with minocycline and rifampin (odds ratio [OR], 0.25 [CI, 0.12 to 0.53]; P = 0.0004); female sex (OR, 0.36 [CI, 0.16 to 0.80]; P = 0.01); systemic use of antibiotics (OR, 0.29 [CI, 0.12 to 0.70]; P = 0.006); insertion into the subclavian vein compared with the jugular vein (OR, 0.39 [CI, 0.18 to 0.84]; P = 0.02) or insertion into the subclavian vein compared with the femoral vein (OR, 0.28 [CI, 0.10 to 0.82]; P = 0.002); and insertion performed outside of the intensive care unit (OR, 0.41 [CI, 0.18 to 0.96]; P = 0.04). Statistical analysis showed no significant interactions among the five factors. Staphylococcus epidermidis alone or with other organisms colonized 25 uncoated and 3 coated catheters (P < 0.001). In addition, gram-positive organisms alone or with other organisms colonized 31 uncoated and 4 coated catheters (P < 0.001). The frequency of colonization of catheters with gram-negative bacilli and Candida species was similar in the two groups (Table 3).

Central Venous Catheter-Related Bloodstream Infection

None of the coated catheters was associated with catheter-related bloodstream infection, but seven uncoated catheters from seven patients led to catheter-related bloodstream infections: Six infections were caused by S. epidermidis, and one was caused by Enterococcus faecalis (P < 0.01, exact log-rank test; risk difference, 5.2% [CI, 0.2% to 12.4%] (Table 3). Each of the five participating hospitals contributed at least one case with catheter-related bloodstream infection. Molecular typing revealed identical DNA patterns of organisms that were isolated from uncoated catheters and blood in four of the catheter-related S. epidermidis bloodstream infections and the E. faecalis bloodstream infection (P = 0.03, exact log-rank test; risk difference, 3.7% [CI, −1.3% to 10.5%]) (Figure 1). The rates of catheter-related bloodstream infection per 1000 catheter-days were 7.34 for uncoated catheters and 0 for coated catheters (P < 0.01, binomial exact test) (Figure 2).

Studies of Antimicrobial Resistance

Fresh catheters coated with minocycline and rifampin showed inhibitory activity against the organisms that were isolated from the indwelling catheters coated with the two drugs. By using the modified Kirby-Bauer technique, the following mean zones of inhibition were determined: 24.3 ± 4.2 mm against S. epidermidis, 26.5 ± 1.5 mm against S. aureus, 6.8 ± 1 mm against C. albicans, and 7.8 ± 4.6 mm against Pseudomonas aeruginosa and other gram-negative bacilli. In addition, the mean zones of inhibition of fresh coated catheters against S. epidermidis isolated from indwelling uncoated and coated catheters were 22.4 ± 5.4 mm and 24.3 ± 4.2 mm (P > 0.2), respectively. The MICs of minocycline and rifampin against all staphylococci (S. epidermidis and S. aureus) isolated from the segments of indwelling catheters coated with minocycline and rifampin and from concurrent insertion sites were 1 µg/mL or less and 2 µg/mL or less, respectively.

Costs Associated with the Use of Catheters Coated with Minocycline and Rifampin

Pittet and colleagues [20] and Heiselman [21] estimated that the extra cost of treating one episode of catheter-related infection in a critically ill patient was $28 690 per survivor (50% of these episodes were caused by coagulase-negative staphylococci) and that such an episode resulted in an additional average stay in the intensive care unit of 6.5 days. Because most bloodstream infections in our study were caused by S. epidermidis, a conservative estimate of cost–benefit was determined by assuming that the cost of a catheter-related bloodstream infection is $14 345 (half the cost reported by Pittet and colleagues and Heiselman [20, 21]). The cost of these infections in the control group was estimated to be $100 415. The cost of catheter-related bloodstream infection per insertion of a control catheter in a single patient is therefore about $738.35. The manufacturer of the catheters coated with minocycline and rifampin expects that these devices would cost $14 more per device than identical catheters not coated with the two drugs (Cook Critical Care, Ellettsville, Indiana. Personal communication). Thus, the cost savings for a critically ill patient who requires a central venous catheter would be $724.35, assuming that all patients with bloodstream infection survive. Of the five hospitals involved in this study, the average annual rate of catheter insertions in critically ill patients is 850 catheters per hospital. Therefore, the use of catheters coated with minocycline and rifampin could save each hospital more than $500 000 per year.

Discussion

In this randomized study, we showed that venous catheters coated with minocycline and rifampin decreased the risk for catheter-related colonization and bloodstream infections. In addition, these antimicrobial-coated catheters were not associated with adverse effects or with the emergence of antimicrobial resistance.

The combination of minocycline and rifampin is unique because both agents are active against methicillin-sensitive and methicillin-resistant staphylococci [22-26] and because both are broad-spectrum agents with reported activity against gram-negative bacilli and Candida species [27-30]. These agents are not antagonistic and, in fact, are occasionally synergistic [31-33]. Kamal and colleagues [8] showed that central venous catheters coated with cefazolin can reduce the risk for catheter-related colonization. Unlike many antimicrobial agents, however, minocycline and rifampin are not routinely used to treat bloodstream infections and thus could be used for prophylaxis with minimal concern about antimicrobial resistance. In an in vitro model of catheter colonization (the modified Robbins device), minocycline plus rifampin was shown to act synergistically to completely prevent the colonization of slime-producing S. epidermidis and S. aureus on catheter surfaces [10]. In this model, the combination of minocycline and rifampin was superior to any other antibiotic combination, including vancomycin plus rifampin, in preventing staphylococcal colonization of venous catheters.

In a randomized clinical trial, Maki and colleagues [9] found that central venous catheters coated with chlorhexidine and silver sulfadiazine may decrease the risk for catheter-related bloodstream infection. In two other studies [10, 11], an in vitro model was used to test the inhibitory activity of antimicrobial catheters (the modified Kirby-Bauer method). Catheters coated with minocycline and rifampin were found to have broad-spectrum inhibitory activity against gram-positive bacteria, gram-negative bacteria, and C. albicans and were significantly superior to catheters coated with chlorhexidine and silver sulfadiazine (as judged by the sizes of the zones of inhibition). In an established rabbit model for catheter infections, catheters coated with minocycline and rifampin were efficacious and were superior to catheters coated with chlorhexidine and silver sulfadiazine in preventing colonization and infection with S. aureus[11].

The results of our study should be interpreted within the framework of the methods and data analysis. The coated catheters that were shown to be efficacious were triple-lumen polyurethane catheters placed through a new percutaneous insertion (not exchanged over a guidewire) for a relatively short period (1 to 28 days). Coated catheters were shown to prevent only catheter-related, gram-positive colonization and bacteremia. This could be related to the small number of catheter-related colonizations and bloodstream infections seen in the control group, most of which were caused by gram-positive bacteremia.

We found no evidence suggesting emergence of resistance to either minocycline or rifampin. The development of resistance is possible but not likely, for several reasons. First, resistance to minocycline and rifampin is most often caused by mutations that occur at a frequency of about 10⁻⁶[34, 35]. One can predict that resistance will most likely emerge in the setting of high bacterial colonization of a specific area, such as the gastrointestinal tract. In our trial, catheters coated with minocycline and rifampin were rarely colonized (8%). Only 5% of these catheters were colonized with at least 10⁴ colony-forming units per catheter segment. Second, neither minocycline nor rifampin was detected in any of the 60 serum samples obtained at various intervals from a cohort of 15 patients with indwelling coated catheters. Thus, the development of bacterial resistance to minocycline and rifampin among organisms residing at sites distant from the catheter is very unlikely.

Third, minocycline and rifampin in combination were shown to be synergistic in preventing the colonization of catheter surfaces [10]. Resistance rarely occurs when a synergistic combination of antimicrobial agents is used. Fourth, superinfections are unlikely because catheters coated with minocycline and rifampin have broad-spectrum inhibitory activity against gram-positive bacteria, gram-negative bacteria, and Candida species [11]. Finally, it is possible to postulate that, by decreasing the risk for catheter-related bloodstream infection, coated catheters may decrease the need for systemic use of such antibiotics as vancomycin and oxacillin. This, in turn, would decrease the likelihood of resistance to therapeutic agents used in the empirical treatment of bloodstream infections. However, continuing surveillance for resistance is needed as coated catheters come into further clinical use.

The recent Centers for Disease Control and Prevention guidelines for the prevention of catheter-related infections recommend the use of antimicrobial-impregnated venous catheters for patients with a high rate of infection after full adherence to other infection control measures, such as maximal sterile barrier precaution [18]. The rate of catheter-related infection associated with triple-lumen polyurethane catheters for short-term use inserted under maximal sterile barrier in our study was higher than the rate associated with mostly single-lumen silicone catheters for long-term use inserted under the same conditions, as described in a previous study [36]. It should be noted that, despite the use of maximal sterile barrier precautions in that study [36], the rate of catheter-related bloodstream infection continued to be 5% in the control group. However, this complication was prevented in our study by the use of catheters coated with minocycline and rifampin. Therefore, this technology should not be a substitute for standard infection control and aseptic measures but rather should be complementary to such efforts.

In conclusion, our study suggests that anti-infective catheters coated with minocycline and rifampin may provide a safe means of reducing the incidence of catheter-related bloodstream infection and decreasing the risk for catheter colonization and may save some hospital costs. Additional randomized studies are required to compare the efficacy of various types of antimicrobial-coated polyurethane catheters used for the short term and silicone catheters used for the long term.

Appendix

The following physicians and hospitals participated in the Texas Medical Center Catheter Study Group: Stephen Greenberg, MD, Nicolas Hanania, MD, and Daniel Yosher, MD (Harris County Hospital District [Ben Taub General Hospital], Houston, Texas); Donald Gibson, MD, Michael Reardon, MD, Patrick Reardon, MD, and Salwa Shenaq, MD (The Methodist Hospital, Houston, Texas); Tukaram Darnule, MD, PhD, and Mohammad Mansouri, BS (Veterans Affairs Medical Center, Houston, Texas); and Kenneth Rolston, MD, Estella Whimbey, MD, Carol Bivins, RN, Armando Huaringa, MD, Kristin Price, MD, and Hossam Safar, MD (M.D. Anderson Cancer Center, Houston, Texas).

Drs. Wall and Robertson: Harris County Hospital District (Ben Taub General Hospital), 1504 Taub Loop, Houston, TX 77030. Drs. Harris, Shenaq, and Curling: The Methodist Hospital, 6565 Fannin Street, Houston, TX 77030.

Dr. Ericsson: The University of Texas Health Science Center, 6411 Fannin Street, Houston, TX 77030.